Abstract: Targeted at the difficulity in measuring objectively users’ emotions in interacting with webs，a method was proposed by using eye movement data. Firstly，the PAD scale in Chinese and the initial eye movement indexes were selected，the representative webs were chosen through a questionnaire，and the interaction tasks were designed. Secondly，the users’ eye movement and PAD data were obtained through an experiment. Then the eye movement indexes related to emotions were determined after the variance homogeneity test and oneway ANOVA，placing emotional classification as the influential factor. Finally，the relational models between eye movement data and users’ emotions were constructed with the partial least square method，and the effectiveness and adaptability of the models were verified. The results indicated that the models have good performances in measuring web users’ emotions.

Key words: eye tracking, emotional measurement, interactive tasks, PAD scale, PLS(partial least square)